Skip to content
Menu

Mahmoud Atieh

AI Automation Engineer

I build AI systemsthat automate real work.

Agents, retrieval systems and automated workflows — built as products that run in production, grounded in real data and connected to real tools.

Mahmoud Samir Atieh
02/Shipped Projects

Shipped Projects

Full-stack AI products I built end to end and deployed live to production.

03/What I build

What I build

Four core systems architected for production and verified in real use.

01 / Autonomous Reasoning
Sub-second tool calls

AI Agents

Production agents connected to real APIs, tools and databases — they resolve intent, call the right service dynamically, and answer from live data instead of scripted flows.

ReAct Tool LoopsDynamic Intent RoutingSchema ValidationAPI Integrations
Architecture SpecProduction Ready
02 / Retrieval & Grounding
100% verifiable citations

RAG Systems

Grounded retrieval pipelines built for precision and zero hallucination: document OCR ingestion, dense vector & BM25 hybrid search, and cross-encoder re-ranking.

Hybrid Vector + BM25pgvector EmbeddingsRe-ranking EngineZero Hallucination
Architecture SpecProduction Ready
03 / Integration & Sync
24/7 background execution

Workflow Automation

Resilient API-driven automations with webhooks, OAuth and scheduled jobs — taking complex manual processes off people's desks with battle-tested error handling.

Self-Hosted n8nWhatsApp & Shopify APIsEvent-Driven QueuesOAuth 2.0 Flow
Architecture SpecProduction Ready
04 / Full-Stack Infrastructure
High-uptime runtime

Production Software

The robust engineering that makes AI features dependable: type-safe backends, relational databases, custom dashboards, containerized deployment, and live monitoring.

TypeScript & PythonPostgreSQL & PrismaDocker & Linux VPSEdge Deployment
Architecture SpecProduction Ready
04/About

About

Building retrieval systems, agents and automated workflows, with 1+ year of experience in AI automation.

I build the product around them too — the API, the database, the dashboard and the deployment. A model that performs in a notebook is a long way from a system someone can depend on at nine in the morning.

I care about grounding. In every project here, an answer traces back to something real — a page in a document, a product record, an order. That constraint shapes the architecture more than the model choice does.

agent_telemetry.log
TASK: Autonomous RAG Retrieval & SynthesisINFERENCE LIVE
[00:01]systemInbound query received: 'Summarize quarterly contract liabilities'tokens: 12
>
Based in
Istanbul, Turkey
Focus
AI agents · Retrieval · Automation
05/Contact

Have a system that needs building? Let's talk.

Send the process as it actually happens — the tools, the volume, what goes wrong. I'll tell you whether it's worth automating and what I'd build first.